Amplifying genomic libraries: How many PFU's per plate?

frist at frist at
Tue Jul 2 16:20:29 EST 1991

In article <0094AFE5.D981A160 at> kim at writes:
>In article <1991Jun27.220747.16895 at>, frist at writes:
>>I am preparing to screen a pea genomic library in Lambda L47.1. With an 
>>average insert size of 20kb, approximately 1.2e6 clones are needed to
>>have a 99% chance of finding a single copy gene.
>                      (stuff omitted)
>>I haven't amplified a genomic library before, but I don't recall hearing
>>about this scale of operation from others who have.
>                       (More omitted)
>First, I am not sure that one can generate adequate representation by simply
>amplifying a library to get more pfu.  I always assumed that the calculated
>minimum pfu for a representative library referred to the size of the library
>before amplification.  That is, a "library" that has initially only 10
>recombinant plaques can be amplified to contain millions of pfu, but still not
>contain any more genetic information.  Does anyone else have thoughts on this?
This is exactly right. In fact, the big worry with amplification is that,
with each round of amplification, competition and even sampling error will
cause the complexity (ie. the amount of genetic information) to decrease.
So ideally, you'd like to always screen with a new, freshly-packaged
library. However, because making libraries de-novo is still very much an
exercise in black magic, people try to get the most out of libraries,
especially those that have performed well in the past. Unfortunately, when
you only have enough PFU's to do ONE plating, the library is not terribly
useful. Anything can go wrong in one plating, and what if you want to
screen for other things later on? Hence, the need to be able to amplify
>Second, one can plate a lambda library at very high densities for amplification
>if the plate lysate is harvested before the plaques are large enough to touch
>each other.  This may mean harvesting at a time before plaques are really
>visible, which requires a leap of faith.
>Daniel Kim                                    
The consensus of those who responded to my original post was that yes, you
can often get away with plating at high density (~1e5 per 150mm plate) but
if you want to maximize the likelihood that the amplified library will be a
representative of the original, you should use low density (~2e4 per
plate). This minimizes competition, as well as decreasing the probability
of multiple infection of the same cell with two phage, which could result
in recombination artifacts.

Many thanks to all who have responded to my question.

Brian Fristensky                |  
Department of Plant Science     | Freedom begins when you tell Mrs. Grundy
University of Manitoba          | to go fly a kite.
Winnipeg, MB R3T 2N2  CANADA    |  
frist at          | 
Office phone:   204-474-6085    |   - Robert A. Heinlein
FAX:            204-275-5128    |

More information about the Methods mailing list